Re: [Swig-user] typemap(out) for Perl and FILE*

[Petr N. <Pet...@am...>]

> From: Petr Novotny=0D
> Sent: 06 November 2006 19:02
> To: swi...@li...
> Subject: typemap(out) for Perl and FILE*
>=0D
> Folks,
>=0D
> before I start reinventing a wheel, I'd like to ask - is=0D
> there a kind soul out there who wrote a %typemap(out) for=0D
> FILE* for Perl and is willing to share it?=0D

I've done some scavenging on the internet and compiled together the=
 following typemap. It seems to work with Perl 5.8.

%typemap(out) FILE* {
  /* first, an empty glob */
  GV *gv =3D (GV*)SvREFCNT_inc(newGVgen("main"));
  hv_delete(GvSTASH(gv), GvNAME(gv), GvNAMELEN(gv), G_DISCARD);

  /* get its IO */
  IO *io=3DGvIOn(gv);

  /* convert the FILE* into PERLIO* */
  PerlIO *perlio=3DPerlIO_importFILE($1,NULL);

  /* put the PERLIO into the glob */
  IoIFP(io)=3Dperlio;
  IoOFP(io)=3Dperlio;

  /* return a reference to the glob */
  ST(argvi++) =3D sv_2mortal(newRV_noinc((SV*)gv));
}

I am sure it contains hardcoded assumptions that are not fully correct=
 (e.g., it doesn't check for gv=3D=3DNULL).

1. Is there some kind soul who can comment on it, and perhaps suggest=
 improvements?
2. Whoever wants to use this typemap: Feel free to use it. Just note that=
 the typemap assumes ownership of the FILE* in a way - for example, you are=
 no longer allowed to fclose() it. (If you need to keep ownership, you may=
 want to fdopen() it if you solve buffering issued.) Also, it doesn't solve=
 test/binary mode of the FILE (I'm on UNIX and therefore I don't have to).


Regards,

--
Petr Novotn=FD
Prague Development Center, AMDOCS
+420266773112, +420608711517
"The power of accurate observation is commonly called cynicism by those who=
 have not got it." - G.B.Shaw

This message and the information contained herein is proprietary and=
 confidential and subject to the Amdocs policy statement,
you may review at http://www.amdocs.com/email_disclaimer.asp